446 



Fishery Bulletin 92|2), 1994 



ENSO variability. In 1987, with "cool upwelling" con- 

 ditions contracted eastward and southward at the 

 equator as part of that year's El Nino, in the south- 

 ern ETP (south of about 2°S) common dolphins oc- 

 curred only in the far east off South America. In 1988 

 during the strong La Nina these conditions were well 

 established along the equator to the western extent 

 of the study area, and common dolphins occurred in 

 equatorial waters as far west as 110°W. 



The maps in Figures 6 and 7 are imprecise repre- 

 sentations of species-environment patterns derived 

 by CCA for two reasons. First, the maps show only 

 presence-absence, while we used an effort-corrected 

 index of abundance (daily encounter rate) in the CCA. 

 Second, the contouring requires some smoothing and 

 interpolation between sites, while the CCA compared 

 abundance indices only to environmental variables 

 measured during the same day, along the same track 

 lines where the cetaceans were sighted. These spe- 

 cies are apparently separating more strongly on a 

 smaller scale than we could effectively represent on 

 the maps. A further consideration is that the scaling 

 of axes for biplot presentation was done by using a 

 method in which the canonical scores (as plotted on 

 the maps here) are rescaled to produce biplot loca- 

 tions (Ter Braak, 1988). The resulting ordination gives 

 an accurate relative placement of species centroids, but 

 does not allow direct projection of centroids or toler- 



ances onto canonical axis values as mapped in Figures 

 6 and 7. 



The small but significant interannual variation in 

 the species data was effectively accounted for by 

 interannual variation in the environment. This was 

 demonstrated by the low eigenvalue (^.=0.02, P=0.06) 

 for interannual differences after extracting variance 

 associated with the six oceanographic variables. This 

 result does not necessarily apply to total population 

 abundances, however, because in the above analy- 

 ses we did not include school size estimates in our 

 species data. 



Group size effects 



Inclusion of group-size data in the dolphin abundance 

 index produced ordinations that were very similar 

 to those using simple encounter rates, but with a 

 slight decrease in explanatory power from the envi- 

 ronmental data. We interpret this result to indicate 

 that schools of all sizes occupy approximately the 

 same habitats, and that school size variability within 

 these habitats is not strongly related to the environ- 

 mental variables analyzed here. 



Applications to dolphin assessments 



We suggest two approaches to use the results of this 

 study in cetacean abundance and trend monitoring. 



